\documentclass[12pt]{article}
\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}
\usepackage{fullpage}
\usepackage{amsmath}
\usepackage{amsthm}
\usepackage{array}
\usepackage{color}
\usepackage{graphicx}
\usepackage{float}
\PassOptionsToPackage{hyphens}{url}
\usepackage[hidelinks]{hyperref}
\usepackage{listings}
\usepackage[margin=1in]{geometry}
\usepackage{setspace}
\usepackage{natbib}
\usepackage{proof}
\usepackage{multirow}
\usepackage{hhline}
\usepackage{wrapfig}
\usepackage [english]{babel}
\usepackage{grffile}
\usepackage{tikz}
\usepackage{pgfplots}
\usepackage[tikz]{bclogo}
\usetikzlibrary{chains}
\usetikzlibrary{positioning}
\usetikzlibrary{arrows}
\usepackage{lscape}
\usepackage [autostyle, english = american]{csquotes}
\usepackage{enumitem}
\usepackage{caption}
\usepackage{subcaption}
\usepackage{indentfirst}
\usepackage{hyperref}
\usepackage{pdfpages}
\usepackage{booktabs}
\newcommand{\tabitem}{~~\llap{\textbullet}~~}
\bibliographystyle{apsr}
\bibpunct{(}{)}{;}{a}{,}{,}
\DeclareGraphicsExtensions{.pdf,.png,.jpg}
\setlength{\tabcolsep}{.18cm}
\usepackage{fancyvrb}
\usepackage{sectsty}
\sectionfont{\fontsize{14}{14}\selectfont}
\subsectionfont{\fontsize{12}{12}\selectfont}
\usepackage{numprint}
\npthousandsep{,}

\usepackage{listings}
\lstset{
basicstyle=\small\ttfamily,
columns=flexible,
breaklines=true
}

% COLOR CODING OUR COMMENTS
\newenvironment{bluetext}{\color{blue}}{\ignorespacesafterend} 

% ==Cross Referencing Different Docs
\usepackage{xr}
\externaldocument{Colombia_SI}

% FOOTNOTE FORMATTING FOR JOP
\usepackage[]{footmisc}
\renewcommand{\footnotelayout}{\setstretch{2}\normalsize}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\title{Left Out: How Political Ideology Affects Support for\\ Migrants in Colombia\footnote{Funding support for this research was provided by the MacMillan Center for International and Area Studies at Yale University and the Center for Human Values at Princeton University.}
\footnote{This research received institutional review board (IRB) approval from 
Princeton University ($\#$8335), UCLA (\#19-001733), and UBC ($\#$H19-03288). 
Our Pre-Analysis Plan was archived in the OSF repository \url{https://osf.io/v2db9/}. 
Replication files are available in the JOP Data Archive on Dataverse (\url{https://dataverse.harvard.edu/dataverse/jop}).The empirical analysis has been successfully replicated by the JOP replication analyst.
}
}

\author{Alisha Holland\thanks{Associate Professor, Government Department, Harvard University, \href{mailto:aholland@fas.harvard.edu}{aholland@fas.harvard.edu}, \href{http://alishaholland.com/about/}{www.alishaholland.com}}
\hspace{1.5cm}
Margaret E. Peters\thanks{Professor, Department of Political Science, UCLA, \href{mailto:mepeters@ucla.edu}{mepeters@ucla.edu}, \href{http://www.maggiepeters.com/}{www.maggiepeters.com}}
\hspace{1.5cm}
Yang-Yang Zhou\thanks{Assistant Professor, Department of Government, Dartmouth College, \href{mailto:yang-yang.zhou@dartmouth.edu}{yang-yang.zhou@dartmouth.edu}, \href{https://www.yangyangzhou.com/}{www.yangyangzhou.com}}
}


\date{\today 
}
%%%%%%%%%%%%%%%%% END OF PREAMBLE %%%%%%%%%%%%%%%%


\begin{document}

<<eval=TRUE,echo=FALSE, results='hide', message=FALSE>>= 
library(knitr)

opts_chunk$set(cache = TRUE, 
        cache.path = 'cache_paper/',
        fig.path = 'figures_paper/', 
        tidy = TRUE, 
        echo = FALSE, 
        warning = FALSE, 
        message = FALSE, 
        fig.pos = 't',
        dev = 'pdf', 
        dpi=200)

options(width = 110, digits = 1, scipen=10000)

@


\maketitle

\begin{abstract} 
\begin{singlespace} 
\noindent 
Do perceptions of migrants' politics affect their reception? The overlap between ethnicity and partisanship often makes it difficult to disentangle the role of political identity. We leverage a case in which migrants come from a similar language and religious background to isolate the role of political perceptions.  We draw on a unique survey of 1,000 Colombian citizens and 1,600 Venezuelan migrants in Colombia to establish the extent of political misperceptions and their effects on migrant reception.  Colombians view Venezuelan migrants as left-wing even though actual Venezuelan migrants are more right-wing than their Colombian hosts. In a conjoint experiment, we find that Colombians oppose the settlement of left-wing migrants in their communities and political views matter more than race, skill, or humanitarian need. These findings point to overlooked political tensions around migration, which may intensify as migrants gain voting rights and politicians reinforce migrant stereotypes for electoral gain.\\   
\\
\textbf{Keywords:} migration, political misperceptions, conjoint experiment, Colombia, Venezuela\\
\end{singlespace} 
\end{abstract}

<<eval=TRUE, echo = FALSE, tidy=TRUE, warning=FALSE, error=FALSE, message=FALSE>>=

setwd("Paper_Inputs")

## Load packages and functions
source("functions.R")

## Load cleaned data
df_col <- readRDS("colombia_clean.RDS") 
df_ven <- readRDS("venezuelans_clean.RDS") 

##Additional Cleaning
### Additional variable transformations

## Colombians
df_col <- df_col %>% # rescale these variables to min 0 max 1 for observational regression analysis
 mutate(dir_contact_bi = case_when(dir_contact_index > 1~1, 
                  dir_contact_index <= 1~0,
                      TRUE~NA_real_),
     open_index_res = scales::rescale(open_index), 
     partisanship_res = scales::rescale(partisanship),
     skilled_labor_res = scales::rescale(skilled_labor),
     contract_res = scales::rescale(contract),
     salary_res = scales::rescale(salary),
     benefits_index_res = scales::rescale(benefits_index),
     dir_contact_index_res = scales::rescale(dir_contact_index),
     indir_contact_index_res = scales::rescale(indir_contact_index),
     cultural_index_res = scales::rescale(cultural_index)
     ) 

df_col_cali <- df_col[df_col$city == "Cali",]
df_col_cucuta <- df_col[df_col$city == "Cúcuta",]

df_col_left <- df_col[df_col$partisanship == 1,]
df_col_right <- df_col[df_col$partisanship != 1,]

df_col_contact <- df_col[df_col$dir_contact_bi == 1,]
df_col_nocontact <- df_col[df_col$dir_contact_bi != 1,]

## Venezuelans
df_ven_cali <- df_ven[df_ven$city == "Cali",]
df_ven_cucuta <- df_ven[df_ven$city == "Cúcuta",]

## Load LAPOP merged data
col.merge <- readRDS("col_merged.RDS") # Colombia
ven.merge <- readRDS("ven_merged.RDS") # Venezuela

@

\pagenumbering{gobble}

\newpage
\pagenumbering{arabic}
\setcounter{page}{1}
\setstretch{2}


Do political views affect how migrants are welcomed in host countries? This question has been largely overlooked in a growing body of research on migrant reception. Migrants to wealthy democracies tend to differ in ethnicity, language, and skill sets. To the extent differences in political views exist, they are hard to disentangle from broader racial, religious, and labor market concerns thought to drive attitudes toward migrants. Yet the largest migrant flows occur between neighboring countries in the Global South \citep{IOM:2018}.  Migrants are often more similar to locals in their demographics but flee governments that pursue extreme ideological projects. Left-wing governments in contemporary Venezuela, North Korea, and Cuba have produced large population outflows, as did right-wing and nationalist dictatorships in El Salvador, Eritrea, Japan, and Nazi Germany.  

Little is known about how citizens in receiving countries perceive migrants' politics or if it matters for their treatment.\footnote{Throughout this article, we use ``migrants'' to refer to a mixed flow of those leaving due to forced displacement crises of all types, while recognizing that \textit{migrants} is not a value-neutral term and affects the legal protections afforded to individuals and their reception.}   On the one hand, host communities may understand that migrants flee out of political opposition, much like political exiles.  Many migrants, and even those that leave for economic or humanitarian reasons, come to oppose the government ideology that forces them to leave \citep{Lim2022}. Many governments allow emigration and may even prefer that political opponents flee abroad \citep{miller2020restraining}. On the other hand, politicians and the media in receiving countries often draw a false equivalence between migrants and the political views of their home governments. Anecdotes abound: Vietnamese refugees fled a Communist regime only to arrive in the United States to be dubbed Communists \citep{flood1977vietnamese}. Syrians are stereotyped as terrorists and ISIL supporters, even though they fled the violence perpetrated by these groups \citep{Rettberga2015, Sisk2016}. Alternatively, the economic vulnerability, discrimination, and precarious legal status of many migrants may lead citizens to see all migrants as natural constituents for Left.  

Political fears can be salient as host citizens worry that migrants will change the electoral dynamics or support extreme political movements in their new homes. Over fifty countries allow noncitizen residents to vote in local, regional, or national elections \citep{ferris2019noncitizen,alarian2021}. Political parties calculate how to appeal to both the existing electorate and migrants that stand to gain voting rights \citep{dancygier2017dilemmas}. Beyond formal rules, citizens in developing countries may worry about informal electoral practices, such as registering and buying off migrant voters before their legal incorporation.

To isolate the role of political concerns, we examine a case where migrants share a language and religious background with their hosts but flee an opposing political context: Venezuelans in Colombia. Venezuelans have migrated in large numbers as the economy collapsed under left-wing populist governments. During the period studied, Colombia was governed by center-right democratic governments. Colombia allows migrants to vote in local elections after five years of residence, which makes Venezuelans' electoral participation a looming issue. We fielded a face-to-face survey with 1,000 Colombians and 1,600 Venezuelan migrants in Colombia before local elections in 2019. The survey is unique in that it includes both Colombians and Venezuelans living in Colombia, allowing us to compare host communities' perceptions to the actual views of migrants.  

We find substantial and consequential political misperceptions. First, a large gap exists between Colombians' political perceptions and Venezuelans' self-reports. While 
\Sexpr{prop.table(table(df_col$pol_left))[2]*100}\% of Colombians believe the majority of Venezuelan migrants identify with the political left, only 
\Sexpr{prop.table(table(df_ven$partisanship))[1]*100}\% of them---even less than Colombians---actually do.  Second, while political misperceptions could mask underlying racial or economic animus, we use a conjoint experiment to disentangle these factors and reduce social desirability bias \citep{horiuchi2020does}. We find that Colombians strongly disfavor migrants from the political Left, and ideology is more important than race or skill in shaping views on migrants. These findings differ from research on attitudes towards immigrants in wealthy democracies that emphasizes racial, cultural, and labor market anxieties.

While we cannot isolate the origins of political misperceptions, we rule out that limited contact or perceived welfare dependence explains misperceptions. We conducted our survey in two cities, Cali and C\'ucuta, that differ in their local demographics. C\'ucuta is a relatively small city on the border, where Venezuelans have long interacted with Colombians. Venezuelans now make up a quarter of the population in some districts. Cali is a large city on the opposite side of Colombia, where Venezuelans are a new addition to the city and constitute less than 1\% of the population. If misperceptions stem from limited exposure to migrants, we should find that individuals living in C\'ucuta with regular contact with Venezuelans or with Venezuelan friends should hold more accurate views of Venezuelans' politics. Instead, we find few differences in political misperceptions based on the city of residence or social contact.  Another possibility is that citizens see migrants as identifying with the political Left due to their economic vulnerability and welfare dependence.  But Colombians oppose the settlement of left-wing Venezuelans even when they are employed and high-skilled. 

Instead, we suggest that voters have concerns about migrants' electoral impact and that national elites cultivate misperceptions for their political advantage. Colombians care more about migrants' political views when they cut against their own, suggesting that citizens are concerned that migrants will swing elections against their preferred candidates.  We draw on examples from Twitter to illustrate how right-wing politicians have fostered misperceptions to discredit moderate left-wing opponents. Colombian politicians stoke fears that Venezuelan migrants support the left, import ``socialist'' ideas, and already vote in elections.  In so doing, our article builds on work on how political entrepreneurs can heighten the salience of some identities over others. While most research focuses on the instrumental creation of ethnic identity \citep[e.g.][]{posner2005institutions,perez2015ricochet}, we emphasize how politicians can successfully leverage citizens' electoral anxieties and use political differences---not just racial or linguistic ones---to mobilize anti-immigrant sentiment and build support at the polls. 

Taken together, our findings highlight the role political misperceptions can play in how migrants are received in host countries. Political fears can be a critical and overlooked driver of hostility towards migrants. Conversely, scholars might overstate the role of ethnic prejudice when it aligns with other political cleavages.   We also provide unique evidence that perceptions of migrants' political views often are wrong. Right-wing politicians have incentives to play up fears of the Left--particularly in countries still shaped by Cold War divides--to mobilize turnout and strengthen their support. 

One implication is that the anticipated electoral consequences of immigration may diverge from those in reality. Despite Colombians' fears that new migrants will push politics to the Left, Venezuelans living in Colombia are even more right-wing than Colombians. Similarly, although the Republican Party in the US warns about ``undocumented Democrats,'' many Latinos fled Communist regimes and hold conservative political views in line with Republicans. Paradoxically, the political incorporation of migrants fleeing left-wing governments can strengthen the very right-wing parties that often decry immigration.

 
\section*{Theory: How Political Identity Shapes Migrant Reception}

\noindent Political concerns are largely absent in the literature exploring public attitudes towards immigration. The underlying assumption is that political identities are less salient than employment, fiscal, or cultural strains from migration.\footnote{For a review, see \citet{hainmueller2014public}; cf. \citet{whitaker2021strategic}.} Yet, many migrants flee extreme ideological projects. The high stakes of political situations that lead to mass emigration can raise concerns in receiving countries about migrants' political loyalties.  About a third of countries extend the right to vote---particularly in local elections---to resident non-citizens \citep{ferris2019noncitizen,alarian2021}, and many migrants can eventually become citizens and gain voting rights. In these contexts, political parties court the votes of migrants \citep{dancygier2017dilemmas}, and the broader public may worry about how migrants will change electoral outcomes.

The political identities of individual migrants cannot be observed by host communities. Hosts need to use heuristics to judge migrants' politics and future electoral impacts.  A common assumption is that migrants share the ideological views of their home governments.  We call this a political false equivalence. This reasoning often is cultivated by the media and politicians who can benefit from associating migrants with their home regime. The Supplementary Information (SI) Section \ref{SIsec:examples} provides numerous historical examples in which anti-immigrant anxieties centered on the political movements in migrants' home countries coming to the host country.\footnote{Supplementary information for this article is available in the appendix in the online edition.}

In much of contemporary Latin America, Venezuelans are stereotyped in the press, on social media, and in political campaigns as socialists or ``Castro-Chavistas.''  The concern is that they will bring leftist ideas with them to the country where they settle.  This logic plays into a sense that migrants transmit the values and politics of their home countries. \citet{Simpser2020} shows that political norms, such as attitudes towards corruption, can be transmitted by immigrants and persist across generations. Immigrants sometimes manifest a stronger attachment to their home country's culture and traditions due to hostility in receiving countries (or ``reactive ethnicity''), which also may extend to the political realm \citep[see also][]{kuo2017social}. Whether propagated by the media, politicians, or popular stereotypes, our expectation is that citizens often see migrants as sympathetic to their home government's politics.  

Concerns about political views may seem to presume well-structured, programmatic politics that do not exist in many developing democracies. Yet politicians may draw ``us'' versus ``them'' distinctions around political groups from foreign countries even when partisanship is weak and poorly defined. The boogeyman of socialism casts a powerful shadow in many developing countries, particularly where neighboring countries experiment with left-wing projects. Host communities also may care about migrants' political impacts because they perceive them as vulnerable to recruitment into armed or criminal groups. Migrants often flee crime and security threats \citep{hiskey2018leaving}, which may raise concerns about importing crime to host countries or acting as ``fifth columns'' \citep{radnitz2022enemies}. Due to the efforts by political entrepreneurs and nearby extreme political projects, ordinary citizens may see migration's potential political impacts as important threats. 

An alternative heuristic is to assume that migrants reject their government's ideology. Cubans in the US, for instance, are loyal Republicans precisely because they experienced Communism and have no desire to see anything like it spread. Forced migrants may be understood as ``voting with their feet'' to leave regimes with which they disagree. Indeed, \citet{Lim2022} shows that European emigrants hold more progressive views and oppose the regime in power more strongly than those who remain. Citizens may assume that migrants--and even those who leave for economic, rather than political, reasons--oppose the government in power in their home country and the ideology that it represents.  

A final possibility is that host communities judge migrants by their class backgrounds.  In contexts in which migrants flee economic and political crises, they often are worse off economically than their host populations.  In this case, citizens may assume that migrants are future constituents for left-wing parties, irrespective of the political context that they left. 

Even if political misperceptions exist, it's unclear whether they matter for migrant reception. Classic explanations of attitudes towards migration emphasize economic factors, such as labor market competition and fiscal strain \citep[e.g.,][]{mayda2006against,hainmueller2015hidden,dancygier2012sectoral,goldstein2014nativism}.  In South-South migration, labor market concerns sometimes intensify because migrants have similar skill sets to host populations \citep[e.g.][]{adida2014immigrant, gaikwad2017majority}. Developing countries also have smaller and lower-quality welfare states, so migration can strain public services that are already insufficient \citep{hanson2007public}. Characterizations that migrants are left-wing may stem from an underlying concern about congestion in social services caused by migration.  The implication is that host communities should be most concerned about the skills and welfare dependence of immigrants, rather than any electoral consequences.  

Immigration attitudes also are often explained by concerns about cultural difference and change \citep[e.g.,][]{bansak2016economic,sides2007european}.  Cultural similarities between immigrants and their hosts may appear greater in developing country contexts, yet group differences still can cause anxiety and rejection by hosts \citep[e.g.][]{thachil2017rural,bhavnani2018nativism,cogley2018which}.  Political misperceptions themselves may be a guise for racial, ethnic, or religious animus towards immigrant groups.  In other words, people may report political misperceptions to justify their anti-immigrant attitudes, particularly if racist views are less socially acceptable. \citet{hopkins2019muted}, for instance, find that Americans overestimate immigrant group size to rationalize their preferences for less immigration.  Political ideology, then, should play a minor role in explaining immigration attitudes once accounting for social desirability bias.
 

\section*{Context: The Politicization of Venezuelan Migration}

Venezuelans and Colombians share many demographic similarities. They speak the same language (Spanish), identify with the same dominant religion (Catholicism), and have mixed skin tones. But the countries have followed divergent political trajectories: Colombia elected right-wing or centrist leaders prior to 2022, while left-wing populist leaders governed Venezuela. 

Colombia has endured more than 50 years of civil conflict involving left-wing guerrilla groups, right-wing paramilitaries, and the state. The conflict killed 220,000 people and displaced an estimated 10\% of Colombia's population \citep{steele2017democracy}. Due to fears of left-wing violence, most Colombians identify with the political Right, and the country didn't follow Latin America's ``left-turn'' in the 2000s. Instead, a right-wing president, \'Alvaro Uribe (2002-10), dominated Colombian politics with an internal security agenda. In 2017, a center-right president, Juan Manuel Santos, struck a fragile peace deal with the country's largest guerrilla group ({\em{Fuerzas Armadas Revolucionarias de Colombia}}, FARC).   

In 2018 and 2022, Gustavo Petro, a former left-wing guerrilla and mayor of Bogot\'a, emerged as the Left's presidential candidate. He won 42\% of the vote in the 2018 run-off and won the presidency in 2022.  He disavowed associations with socialism and violence, instead running on a left-wing platform to support the peace process and attend to Colombia's marginalized groups. While some Colombians still associate the Left with Communism and armed struggle, Petro's victory solidified a Left committed to electoral democracy and social welfare provision.

Colombia's neighbor, Venezuela, has elected left-wing populist presidents since 1999. President Hugo Ch\'avez embarked upon ``21st-century socialism,'' which included nationalizing the country's oil company, implementing price controls, and expanding social spending. While Ch\'avez enjoyed broad popular support, an economic crisis ensued as oil prices and production fell, and increased repression and electoral fraud diminished support for his successor, Nicol\'as Maduro. Venezuela's GDP has shrunk by 62\% since Maduro took office in 2013, one of the largest economic collapses outside of war. For many Venezuelans, the political Left now is associated with authoritarianism, state economic management, and shortages of essential goods.

Venezuela's economic and political turmoil has forced broad swaths of the population to leave. Colombia has received the largest number of Venezuelans, with 1.8 million as of 2019. While early migrants from Venezuela tended to be political or economic elites, recent migrants largely fled the country's economic collapse. The Colombian press and government widely refer to Venezuelans as migrants, rather than refugees. Few Venezuelans have filed formal asylum claims or received refugee status. 

The Colombian government largely has tried to integrate Venezuelans into the labor market and social programs.  In 2017, Colombia created special residency permits ({\em{Permiso Especial de Permanencia, PEP}}) that granted Venezuelans two years of legal residency and access to work permits, education, and public health care.\footnote{For an overview of Colombia's response, see ``Todo lo que tiene que saber sobre la migracion venezolana,'' Ministerio de Relaciones Exteriores, November 1, 2018, \url{www.migracioncolombia.gov.co/infografias/todo-lo-que-tiene-que-saber-sobre-la-migracion-venezolana}}  The government expanded the program to all Venezuelans in 2021. Venezuelans--like all foreigners--will be able to vote in local elections in Colombia after five years of residency.\footnote{Law 1070 of 2006.} Given that the increase in Venezuelan migration began in 2015, most Venezuelans could not vote in the 2018 presidential and 2019 mayoral elections, which are the focus here. 

Colombia differs from many advanced industrial economies in that right-wing politicians led the welcoming response to migrants. President Santos spearheaded the initial tolerant response to Venezuelans, drawing on a long history of Venezuela receiving migrants from Colombia's civil war. Subsequently, President Iv\'an Duque (2018-22), who also came from the political Right, regularized the status of Venezuelan migrants, emphasizing the shared history and solidarity between the neighboring countries.\footnote{``Duque presenta proyecto para regularizar a los migrantes venezolanos, de qu\'e trata?,'' {\em{El Espectador}}, February 8, 2021.}  While Colombian leaders drew on shared migration history, the lack of a xenophobic response is not unique---Brazil, Chile, and Peru all welcomed Venezuelans under right-wing administrations and most provide paths to citizenship with full voting rights \citep{escobar2017migration}.  

The Colombian government has attempted to prevent the politicization of migration at the local level. Before the 2019 local elections that form the backdrop for our survey, the Attorney's General Office led a campaign to encourage tolerance towards Venezuelan migrants. The main political parties made a pact not to campaign on anti-migrant platforms.\footnote{``Partidos colombianos firman pacto pol\'itico contra la xenofobia,'' \textit{Semana}, 22 March 2019, \url{migravenezuela.com/web/articulo/pacto-politico-contra-la-xenofobia-en-elecciones-2019-1015}.} With some important exceptions, local campaigns involved little explicit discussion of migration issues \citep{Woldemikael2022}.

Nonetheless, right-wing politicians leveraged fears of a Venezuela-style economic collapse to their advantage.  In the polarized 2018 presidential race, politicians tried to tie Venezuelan migrants to support for leftist \textit{economic} policies. In particular, former president Uribe used his social media presence to scare voters that Venezuelans were bringing left-wing ideas of ``castrochavismo,'' would vote for the left, and create a ``second Venezuela.'' Rumors circulated that Venezuelans could vote in the election.\footnote{See the fact-checking site, ``No es cierto que todos los venezolanos puedan votar en elecciones de Colombia,'' Colombia Check, May 14, 2019.} 

While the attempt to associate Venezuelan migrants with the Left is most strongly associated with Uribe, he was not alone. President Duque used Venezuela to discredit his opponent, Petro, as a radical. Propaganda for Duque advertised, ``The tragedy of Venezuela is the result of a socialist government. Vote wisely, vote Duque'' \citep{Ordonez2019}. A leading center-right presidential candidate, Germ\'an Vargas Lleras, also wrote, ``It's worrying to think that the tragedy in Venezuela can repeat itself in Colombia. I propose to stop it!''\footnote{``Sobre Eln y Venezuela, hay que poner orden ya,'' {\em{El Tiempo}}, February 18, 2018; \url{www.eltiempo.com/opinion/columnistas/german-vargas-lleras/sobre-eln-y-venezuela-hay-que-poner-orden-ya-german-vargas-lleras-184028}} We return to these maneuvers and their role in fostering political misperceptions in the mechanism section. 

Evidence suggests that the political Right and independent politicians have benefited from increased migration. \citet{rozo2021brothers} show that Colombian voters in municipalities that received more Venezuelan migrants were more likely to turn out and elect right-wing candidates. \citet{Woldemikael2022} finds that increased migration resulted in the entrance and victory of more independent candidates at the local level. Yet such studies of electoral impacts can't distinguish whether migrant-receiving communities are concerned about the political effects of migration or more common cultural, fiscal, and security anxieties.     


\section*{Data and Research Design}

\subsection*{Case Selection and Survey Sampling}

We conducted a face-to-face survey of Colombians and Venezuelan migrants living in Colombia. We timed our survey to coincide with the run-up to local and regional elections in October 2019. Given that political fears generally are not identified as important, we were interested in the most-likely period in which they may matter. 

We surveyed in two Colombian cities, C\'ucuta and Cali, to vary the concentration of migrants. C\'ucuta is located on the Venezuelan border and has absorbed the largest fraction of Venezuelan migrants. Registered migrants constitute 6.34\% of the city population, and the number of unregistered Venezuelans likely is far larger. Residents in C\'ucuta have a long history of interacting with Venezuelans as populations on both sides flow across the border for work and services. In contrast, Cali is Colombia's third-largest city, with smaller concentrations of Venezuelans (less than 1\% of the city population) and little history of interaction. Cali, instead, has received substantial internally displaced people (IDPs) due to its proximity to rural areas with active conflict. Additionally, Cali has a sizable Black population (roughly 20 to 25\%, according to the 2015 census) compared to less than 3\% identifying as Black in C\'ucuta. The comparison of close to ``most-different'' cities (in terms of their experience with Venezuelans) allows us to examine whether differences in contact and local electoral demographics drive political misperceptions. 

We also selected cities to improve the survey quality. Venezuelans in C\'ucuta and Cali are spatially concentrated, making them easier to survey without a standard sampling frame than larger cities (like Bogot\'a and Medell\'in) or rural areas with diffuse migrant populations. 

In each city, we surveyed about 800 Venezuelans and 500 Colombians. We excluded Colombian-Venezuelan dual citizens, given that they likely have different access to services and views on migration.\footnote{Power calculations and funding determined sample sizes. The margin of error for the survey of Colombians in both cities is about 4.5\% and 3.4\% for Venezuelans. In the conjoint experiment, we can find a minimum effect size of 0.1 with a 95\% confidence interval.} The demographics of Colombians and Venezuelans are similar to available surveys and administrative data (see SI Section \ref{SIsec:demog}). Compared to nationally representative samples, our Colombian respondents tended to identify slightly more with the political Left (although most still lean right) and were wealthier, reflecting our urban sample (SI Section \ref{SIsec:lapop}). 

The Venezuelan migrant respondents (N=\numprint{\Sexpr{nrow(df_ven)}}) were younger and more educated than the Colombians on average (\Sexpr{mean(df_ven$education>1, na.rm=T)*100}\% have at least a secondary education). Despite higher levels of education, the vast majority worked in the informal economy in Colombia (\Sexpr{mean(df_ven$contract==1, na.rm=T)*100}\%).  Most left Venezuela for economic reasons (\Sexpr{mean(df_ven$reason_leave_num1==1, na.rm=T)*100}\%). Compared to the last LAPOP survey on Venezuela (2016-2017), our respondents in Colombia are younger, more likely to be looking for work, and have trouble making ends meet, as to be expected of migrants. 

\subsection*{Survey Implementation \& Research Ethics}

We ran a face-to-face survey with a respected Colombian survey firm (Centro Nacional de Consultor\'ia, CNC). Respondents received roughly \$5 USD in compensation to complete the survey, which was equal to about 3 hours at minimum wage. This amount was chosen to compensate people for their time without inducing participation. The survey for Venezuelans lasted approximately one hour; the survey for Colombians lasted 45 minutes. All enumerators were Colombians from the city of the survey.\footnote{Contact information was collected for quality control purposes; more than 10\% of surveyed individuals were re-contacted in each city. Data were kept on a secure server to protect respondents. We programmed the survey in the Qualtrics offline application.} 

<<surveylocal, eval = FALSE, echo = FALSE, tidy=TRUE, fig.width = 8, fig.height = 5, out.width= ".9\\linewidth", fig.align='center', warning=FALSE, message=FALSE, strip.white=TRUE, fig.cap="Maps of Cali (right) in southwest Colombia, and Cúcuta (left) in northern Colombia along the Venezuelan border, showing the locations of our survey respondents. Google basemap.">>=

## Make maps of Colombian and Venezuelan Survey Respondents
df_map <- df_col %>% dplyr::select(longitude, latitude, nationality, city) %>%
  bind_rows(df_ven %>% dplyr::select(longitude, latitude, nationality, city)) %>%
  mutate(longitude = as.numeric(longitude),
      latitude = as.numeric(latitude))

## Get maps from Google - requires api

# Cali
register_google(key = "AIzaSyCOOx1tGoCJJbBOUeH0oSC5C5bNamOd7iI")
cali_map <- get_googlemap(center = c(lon= -76.51, lat=3.425),
           zoom = 12, 
     maptype = "roadmap", 
     color = "bw",
     #api_key = "AIzaSyCOOx1tGoCJJbBOUeH0oSC5C5bNamOd7iI",
     force = TRUE) %>% 
  ggmap() +
  geom_point(data = df_map[df_map$city == "Cali" & 
                is.na(df_map$longitude) == FALSE,], 
             aes(x = longitude, 
               y = latitude, 
               colour = nationality)) +
  scale_colour_manual(values = c("#D55E00","#0072B2"),
            labels=c("Colombians", "Venezuelans")) +
  ggtitle("Cali") +
  theme(axis.line = element_blank(),
    axis.text = element_blank(),
    axis.ticks = element_blank(),
    plot.margin = unit(c(0, 0, -1, -1), 'lines'),
    legend.title = element_blank(),
    legend.justification=c(1,0), 
    legend.position=c(1,0),
    legend.key = element_rect(fill = "transparent", colour = "transparent"),
    legend.background = element_rect(fill=alpha('white', .4))) +
 xlab('') +
 ylab('') 

# Cucuta
cucuta_map <- get_googlemap(center = c(lon= -72.5, lat=7.9),
           zoom = 13, 
     maptype = "roadmap", 
     color = "bw",
     #api_key = "AIzaSyCOOx1tGoCJJbBOUeH0oSC5C5bNamOd7iI",
     force = TRUE) %>% 
  ggmap() +
  geom_point(data = df_map[df_map$city == "Cúcuta" & 
                is.na(df_map$longitude) == FALSE,], 
             aes(x = longitude, 
               y = latitude, 
               colour = nationality)) +
  scale_colour_manual(values = c("#D55E00","#0072B2"),
            labels=c("Colombians", "Venezuelans")) +
  ggtitle("Cúcuta") +
  theme(axis.line = element_blank(),
    axis.text = element_blank(),
    axis.ticks = element_blank(),
    plot.margin = unit(c(0, 0, -1, -1), 'lines'),
    legend.title = element_blank(),
    legend.justification=c(1,0), 
    legend.position=c(1,0),
    legend.key = element_rect(fill = "transparent", colour = "transparent"),
    legend.background = element_rect(fill=alpha('white', .4))) +
 xlab('') +
 ylab('') 

#plot
cali_map + cucuta_map


@

\begin{figure}
\centering
\includegraphics[scale=.35]{surveylocal.png}
\caption{Maps of Cali (right) in western Colombia and C\'ucuta (left) in eastern Colombia along the Venezuelan border, showing the locations of survey respondents. Google basemap.}
\label{fig:surveylocal}
\end{figure}

For Colombian respondents, we conducted a household survey. The survey firm designed a stratified sample according to socioeconomic blocks. Enumerators were assigned a random start location on each block and used a skip rule to select houses. For Venezuelans, we used two strategies to create a high-quality sample due to the lack of an existing sampling frame. First, we conducted a household survey in neighborhoods that official registries identify as having high concentrations of Venezuelans. Enumerators randomly selected blocks as starting points and then used a skip rule between units. However, migrants living in fixed households may be wealthier or live in enclaves with less direct contact with Colombians. As a second strategy, we created a list of places where Venezuelan gather (government offices, spot employment markets, shelters, churches) and randomly selected gathering points. Enumerators then interviewed migrants they met at these points according to a skip rule, finding space to conduct the survey privately. We screened for Venezuelans who had moved in the last three years to create a comparable group of migrants. Figure \ref{fig:surveylocal} shows the locations of our surveys. The Colombian samples are geographically spread across these two cities. The Venezuelans, however, cluster together in enclaves, similar to migrant populations worldwide. 

Our survey, discussed below, focused on opinions about politics and migration. We framed sensitive questions, such as those on recruitment by gangs or irregular groups, in terms of family experiences to improve reporting and avoid recounting personally traumatic experiences. During the consent process, enumerators also ensured that respondents understood that participation was voluntary, they could terminate the survey at any time, and  they did not have to answer every question. We randomized the order of the observational questions and conjoint experiment to minimize order effects. 

\subsection*{Empirical Strategy and Hypotheses}

We use a combination of descriptive and experimental evidence to test the extent and role of political misperceptions in attitudes toward migrants. First, we examine the extent and nature of misperceptions. We leverage our unique survey that included both Colombians and Venezuelan migrants living in Colombia to compare perceived and actual views. If Colombians rely on a false political equivalence, then we expect Colombian respondents to have widespread  misperceptions of Venezuelan migrants. Most importantly, we expect that Colombians \textit{believe} Venezuelans identify with the political left, while most Venezuelans disavow the left. Given that survey respondents often misjudge percentages, we asked Colombian respondents whether they thought the \textit{majority} of Venezuelan migrants had a given set of political views or experiences. We asked Venezuelans directly whether they (or, in the case of sensitive questions, a family member) held a given view.

Second, we included a forced choice conjoint experiment. Respondents evaluated different profiles of migrants and selected which migrant they would prefer to stay in their city and receive housing and employment assistance. A conjoint experiment allows us to compare political views to other factors theorized to influence preferences over migrants, like skills, welfare dependence, and race. Conjoint experiments also minimize social desirability bias as respondents don't need to justify their choice of profiles. We vary the political leanings of the migrant rather than relying on respondents' stereotypes. Each respondent saw five pairs of migrant profiles that randomized the following:

\begin{singlespace}
\vspace{-.6cm}
\texttt{
\small{
\begin{itemize}
\setlength{\itemsep}{-7pt}
  \item Political ideology: left, center, or right
  \item Origin: IDP from Valle del Cauca, IDP from Norte de Santander, or Venezuelan
  \item Skill level: low, medium, or high 
  \item Employment prospects: very likely to get work, unlikely to get work, cannot work
  \item Reason for leaving: Fear of arrest by the government, violence by \\
  guerrillas/irregular forces/paramilitaries, fear of crime in their area, or poverty
  \item Race: Mestizo or Black
  \item Gender: Male or Female
\end{itemize}
}
}
\end{singlespace}

If political ideology shapes migrant reception, we expect Colombians to be less likely to welcome migrants with left-wing political views.  On the other hand, if political misperceptions are adopted to mask other anxieties or because Colombians perceive Venezuelans as dependent on the welfare state, then factors like race, employability, and skill should play a larger role. To test this, we first calculate the Average Marginal Component Effects (AMCEs), which represents the increase in the probability that a profile is chosen if the component value is changed from one level to the other, averaged over all the other attributes. 

While we cannot directly test the origins of political misperceptions and reasons that they matter for migrant reception, we examine plausible channels rooted in direct contact, economic vulnerability, and electoral concerns. First, we asked Colombian respondents how many Venezuelan friends or work colleagues they have. If contact dispels misperceptions, we expect individuals with more close contacts or who live in cities with more experience with Venezuelans, such as C\'ucuta, to have more accurate views of Venezuelans. 

Second, we examine if Colombians' political perceptions reflect welfare stereotypes and fiscal anxieties. If Colombians rely on a welfare heuristic, then we expect to see a correlation between beliefs that Venezuelans support the Left and perceptions that they rely on the welfare state. We ask Colombian respondents if they believe that Venezuelans (a) make it more difficult to access social services, and (b) will lead their taxes to go up.  In the conjoint, respondents should be less willing to accept migrants unlikely to find work and fleeing poverty in Venezuela. If political fears actually reflect underlying fiscal concerns, then political ideology should matter little for migrants who employable and skilled. 

Third, we examine the role of strategic electoral concerns. Citizens should be least likely to favor migrants with opposing political views. Additionally, individuals who believe Venezuelans can vote in elections and live in cities with high electoral concentrations may care more about migrants' political ideology. We estimate conditional conjoint effects based on a respondent's ideology, belief about whether Venezuelans can vote, and city of residence. Finally, we examine the role of political elites through qualitative analysis of media and Twitter messages in the run-up to elections. We look for evidence that right-wing politicians cultivate generalized fears about the electoral participation of Venezuelans.  


\section*{Main Results}

\subsection*{The Extent of Political Misperceptions}

\noindent Our unique survey allows us to compare Colombians' perceptions to Venezuelan migrants' actual preferences and experiences.  Colombians have substantial misperceptions:  
\Sexpr{mean(df_col$pol_left==1, na.rm=T)*100}\% of Colombians think that most Venezuelan migrants support the political left. Additionally, 
\Sexpr{mean(df_col$pol_vote2==1, na.rm=T)*100}\% believe that most Venezuelans already have the right to vote in local elections (even though only
\Sexpr{mean(df_ven$ben_vote==1, na.rm=T)*100}\% have met the five-year residency requirement), and 
\Sexpr{mean(df_col$pol_vote1==1, na.rm=T)*100}\% believe that migrants have the right to vote in national elections, which is reserved for citizens.  Colombians also saw Venezuelans as susceptible to vote buying and armed recruitment. Furthermore, Colombians voiced a variety of fears about Venezuelan migration, from fiscal strain to cultural anxieties (see SI Section \ref{SIsec:colatts}). 

%% TABLE OF MISPERCEPTIONS
\begin{table}[t]
\small{
\begin{tabular}{|>{\raggedright\arraybackslash}p{4cm}|>{\raggedright\arraybackslash}p{5cm}|>{\raggedright\arraybackslash}p{7cm}|}
\hline
      & \textbf{\% of Colombians who believe that the \textit{majority} of Venezuelan migrants}
      & \textbf{\% of Venezuelan migrants} \\\hline
      
Support the left        
& \Sexpr{mean(df_col$pol_left==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_col$pol_left==1, na.rm=T)) * (1 - (mean(df_col$pol_left==1, na.rm=T))) / nrow(df_col[is.na(df_col$pol_left)==F,]) * qnorm(.975))*100})      
& \Sexpr{mean(df_ven$partisanship==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_ven$partisanship==1, na.rm=T)) * (1 - (mean(df_ven$partisanship==1, na.rm=T))) / nrow(df_ven[is.na(df_ven$partisanship)==F,]) * qnorm(.975))*100})             
\\\hline

Support Maduro        
& \Sexpr{mean(df_col$pol_maduro==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_col$pol_maduro==1, na.rm=T)) * (1 - (mean(df_col$pol_maduro==1, na.rm=T))) / nrow(df_col[is.na(df_col$pol_maduro)==F,]) * qnorm(.975))*100})       
& \Sexpr{mean(df_ven$pres_ven==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_ven$pres_ven==1, na.rm=T)) * (1 - (mean(df_ven$pres_ven==1, na.rm=T))) / nrow(df_ven[is.na(df_ven$pres_ven)==F,]) * qnorm(.975))*100})          
\\\hline

Can vote in national elections 
& \Sexpr{mean(df_col$pol_vote1==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_col$pol_vote1==1, na.rm=T)) * (1 - (mean(df_col$pol_vote1==1, na.rm=T))) / nrow(df_col[is.na(df_col$pol_vote1)==F,]) * qnorm(.975))*100})
& 0\% 
\\\hline

Can vote in local elections  
& \Sexpr{mean(df_col$pol_vote2==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_col$pol_vote2==1, na.rm=T)) * (1 - (mean(df_col$pol_vote2==1, na.rm=T))) / nrow(df_col[is.na(df_col$pol_vote2)==F,]) * qnorm(.975))*100})                    
& \Sexpr{mean(df_ven$ben_vote==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_ven$ben_vote==1, na.rm=T)) * (1 - (mean(df_ven$ben_vote==1, na.rm=T))) / nrow(df_ven[is.na(df_ven$ben_vote)==F,]) * qnorm(.975))*100})             
\\\hline

Received a vote-buying offer from Colombian politicians
& \Sexpr{mean(df_col$votebuy_ven==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_col$votebuy_ven==1, na.rm=T)) * (1 - (mean(df_col$votebuy_ven==1, na.rm=T))) / nrow(df_col[is.na(df_col$votebuy_ven)==F,]) * qnorm(.975))*100})                             
& \Sexpr{mean(df_ven$votebuy_col==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_ven$votebuy_col==1, na.rm=T)) * (1 - (mean(df_ven$votebuy_col==1, na.rm=T))) / nrow(df_ven[is.na(df_ven$votebuy_col)==F,]) * qnorm(.975))*100})         
\\\hline

Support guerrilla groups       
& \Sexpr{mean(df_col$pol_guerilla==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_col$pol_guerilla==1, na.rm=T)) * (1 - (mean(df_col$pol_guerilla==1, na.rm=T))) / nrow(df_col[is.na(df_col$pol_guerilla)==F,]) * qnorm(.975))*100})                             
& \Sexpr{mean(df_ven$xeno_guerilla==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_ven$xeno_guerilla==1, na.rm=T)) * (1 - (mean(df_ven$xeno_guerilla==1, na.rm=T))) / nrow(df_ven[is.na(df_ven$xeno_guerilla)==F,]) * qnorm(.975))*100}) had a family member approached by a guerrilla group for recruitment         
\\\hline

Have criminal ties to gangs 
& \Sexpr{mean(df_col$sec_crime==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_col$sec_crime==1, na.rm=T)) * (1 - (mean(df_col$sec_crime==1, na.rm=T))) / nrow(df_col[is.na(df_col$sec_crime)==F,]) * qnorm(.975))*100})                              
& \Sexpr{mean(df_ven$xeno_gang==1, na.rm=T)*100}\% 
($\pm$\Sexpr{sqrt((mean(df_ven$xeno_gang==1, na.rm=T)) * (1 - (mean(df_ven$xeno_gang==1, na.rm=T))) / nrow(df_ven[is.na(df_ven$xeno_gang)==F,]) * qnorm(.975))*100}) had a family member approached by a gang for recruitment
\\\hline
\end{tabular}
}
\caption{The Prevalence of Political Misperceptions: Comparing Colombians' Beliefs with Venezuelans' Reported Experiences. Margins of error in parentheses.}
\label{tab:misperceptions}
\end{table}

In contrast, we find that only 
\Sexpr{mean(df_ven$partisanship==1, na.rm=T)*100}\% of Venezuelans placed their political views on the Left, comparable to the share of Colombians, \Sexpr{mean(df_col$partisanship==1, na.rm=T)*100}\%.\footnote{Non-response is a common problem in measuring political identification. 
\Sexpr{mean(is.na(df_ven$partisanship)==TRUE)*100}\% of Venezuelan migrants did not answer. Even if all of these respondents supported the left, it would still be a minority.} If anything, Colombians are more likely to be centrist and Venezuelans are more likely to identify with the far right. Furthermore, only 
\Sexpr{mean(df_ven$votebuy_col==1, na.rm=T)*100}\% of Venezuelan migrants said that they received an offer to sell their vote and a small minority said that they or a family member were approached to join a guerrilla group (\Sexpr{mean(df_ven$xeno_guerilla==1, na.rm=T)*100}\%)
or a gang 
(\Sexpr{mean(df_ven$xeno_gang==1, na.rm=T)*100}\%).

It is possible that the limited identification of Venezuelans with the political Left reflects an attempt to disavow elite portrayals of them as leftists. However, 
\Sexpr{nrow(df_ven[df_ven$chavez_vote == 1 & is.na(df_ven$chavez_vote) == F,])/(nrow(df_ven[df_ven$chavez_vote %in% c(1,2) & is.na(df_ven$chavez_vote) == F,]))*100}\% 
of Venezuelans admitted that they voted for President Hugo Ch\'avez in the past and \Sexpr{(prop.table(table(df_ven$chavez_rate))[4]+prop.table(table(df_ven$chavez_rate))[5])*100}\% evaluated his presidency as ``good'' or ``very good'' on a 5-point scale. 
This suggests that Venezuelans are willing to admit left-wing political behaviors. Crucially, many no longer identify with the Left. While 
\Sexpr{mean(df_ven$maduro_vote==1, na.rm=T)*100}\% had voted for Maduro, only two respondents 
(\Sexpr{mean(df_ven$pres_ven==1, na.rm=T)*100}\%) said that they wanted Maduro to continue as president and 
\Sexpr{(prop.table(table(df_ven$maduro_rate))[1]+prop.table(table(df_ven$maduro_rate))[2])*100}\% viewed his presidency as ``bad'' or ``very bad'' on a 5-point scale. 

The left-right scale also may be different for Venezuelans, given the increasingly authoritarian and socialist nature of the Venezuelan Left.  In substance, then, a Venezuelan who identifies with the Right may hold similar views to a Colombian who positions herself on the Left.  Yet, a substantive turn away from the left seems apparent among Venezuelans.  For instance, when asked who they would vote for in the 2018 Colombian presidential elections (if they could participate), 
\Sexpr{mean(df_ven$pres_vote_col==1, na.rm=T)*100}\% of Venezuelans said that they would support Duque, which far exceeds the fraction supporting Petro (\Sexpr{mean(df_ven$pres_vote_col==2, na.rm=T)*100}\%). 
 
 
\subsection*{The Importance of Political Ideology}

<<main_conj, eval = TRUE, echo = FALSE, tidy=TRUE, fig.width = 7.5, fig.height = 6.5, out.width= ".9\\linewidth", fig.align='left', fig.pos='t', warning=FALSE, message=FALSE, strip.white=TRUE, fig.cap="Conjoint estimates with 95\\% CIs for Colombian respondents asked to choose their preferred migrant profile in a pairwise comparison.">>=


#### Colombians

## Full conjoint dataframe
df_cjoint_col <- df_col %>% 
 dplyr::select(matches("con[1-5]"), 
        starts_with("rd_"),
        partisanship,
        skilled_labor,
        salary,
        benefits_bi,
        indir_contact_index_2,
        city,
        pol_vote1) %>% 
 dplyr::mutate(id = 1:n(),
        partisanship = as.factor(partisanship),
        skilled_labor = as.factor(skilled_labor),
        salary = as.factor(salary),
        benefits_bi = as.factor(benefits_bi),
        indir_contact_index_2 = as.factor(indir_contact_index_2),
        city = as.factor(city),
        pol_vote1 = as.factor(pol_vote1))

## Clean responses
df_cjoint_response_col <- df_cjoint_col %>%
 dplyr::select(id, matches("con[1-5]")) %>%
 gather(key = round, value = profile_choice, -id) %>%
 mutate(round = parse_number(round))

## Clean attributes
df_cjoint_covs_col <- df_cjoint_col %>%
 dplyr::select(id, starts_with("rd_")) %>%
 gather(key = feature_round_profile, value = attribute, -id) %>%
 mutate(round = parse_number(feature_round_profile),
     profile = case_when(grepl("_a_", feature_round_profile)~"a",
               grepl("_b_", feature_round_profile)~"b",
               TRUE~"ruh-roh"),
     feature = gsub("rd_[1-5]_[a|b]_", "", feature_round_profile)) %>%
 dplyr::select(-feature_round_profile) %>%
 spread(key = feature, value = attribute)

## Bind responses and attributes together
df_cjoint_clean_col <- inner_join(df_cjoint_response_col, df_cjoint_covs_col,
               by = c("id", "round")) %>%
 left_join(df_cjoint_col %>% dplyr::select(id, 
                      partisanship,
                      skilled_labor,
                      salary,
                      benefits_bi,
                      indir_contact_index_2,
                      city,
                      pol_vote1)) %>%
 mutate(chosen_profile = case_when(profile_choice == 1 & profile == "a"~1,
                  profile_choice == 2 & profile == "b"~1,
                  profile_choice == 1 & profile == "b"~0,
                  profile_choice == 2 & profile == "a"~0,
                  TRUE~NA_real_),
     # Partisanship = as.factor(Partisanship), 
     # `Skill Level` = as.factor(`Skill Level`), 
     # `Identity of Migrant` = as.factor(`Identity of Migrant`),
     # `Probability of Employment` = as.factor(`Probability of Employment`), 
     #  Race = as.factor(Race), 
     # `Reason for Leaving` = as.factor(`Reason for Leaving`), 
     # Gender = as.factor(Gender)
     PARTISANSHIP = as.factor(Partisanship), 
     `SKILL LEVEL` = as.factor(`Skill Level`), 
     `IDENTITY OF MIGRANT` = as.factor(`Identity of Migrant`),
     `PROBABILITY OF EMPLOYMENT` = as.factor(`Probability of Employment`), 
      RACE = as.factor(Race), 
     `REASON FOR LEAVING` = as.factor(`Reason for Leaving`), 
     GENDER = as.factor(Gender)
     ) %>%
 drop_na(-c(partisanship,
        skilled_labor,
        salary,
        benefits_bi,
        indir_contact_index_2,
        city)) %>%
 mutate(pair_id = paste0(id, round)) %>%
 group_by(pair_id) %>%
 filter(n() == 2) %>% ungroup()

## Run conjoint estimator
set_baseline <- list(PARTISANSHIP = "Left",
           `SKILL LEVEL` = "Low",
           `IDENTITY OF MIGRANT` = "Colombians displaced from around Cali",
           `PROBABILITY OF EMPLOYMENT` = "Cannot work",
            RACE = "Mestizo",
           `REASON FOR LEAVING` = "Fear of crime in their area")

df_cjoint_clean_col$PARTISANSHIP <- fct_relevel(df_cjoint_clean_col$PARTISANSHIP, "Left", "Center", "Right")
df_cjoint_clean_col$`SKILL LEVEL` <- fct_relevel(df_cjoint_clean_col$`SKILL LEVEL`, "Low", "Medium", "High")

amce_out.col <- cjoint::amce(chosen_profile ~ PARTISANSHIP + `IDENTITY OF MIGRANT` + `SKILL LEVEL` +
    `PROBABILITY OF EMPLOYMENT` +`REASON FOR LEAVING` + RACE + GENDER,
   data = df_cjoint_clean_col, 
   respondent.id = "id", 
   baselines = set_baseline)

amce_out.col$estimates <- lapply(amce_out.col$estimates, function(x){
 est <- x[1,]
 stderr <- x[2,]
 ci_low <- est - stderr * qnorm(.975)
 ci_hi <- est + stderr * qnorm(.975)
 xout <- rbind(est, stderr, ci_low, ci_hi)
 colnames(xout) <- colnames(x)
 return(xout)
})

## Plot results
cbPalette <- c("#99098d", "#000000", "#D55E00", "#009E73", "#E69F00", "#999999", "#0072B2")

amce_out.col.plot <- cjoint_plot(amce_out.col, 
   xlim=c(-.27,.18), breaks=c(-.2, -.1, 0, .1), labels=c("-.2", "-.1", "0",".1"), 
   xlab="Change in Pr(Preferred Migrant)",
   #main="Colombian respondents",
   lwd = 1.5,
   colors = cbPalette,
   group.order = c("PARTISANSHIP", 
                   "IDENTITY OF MIGRANT", 
                   "SKILL LEVEL",
                   "PROBABILITY OF EMPLOYMENT", 
                   "REASON FOR LEAVING", 
                   "RACE", 
                   "GENDER")) +
   theme_bw() + 
   theme(legend.position = "none") 

#### Venezuelans

## Full conjoint dataframe
df_cjoint_ven <- df_ven %>% 
 dplyr::select(matches("con[1-5]"), 
        starts_with("rd_"),
        partisanship,
        skilled_labor,
        salary,
        benefits_bi,
        city) %>% 
 dplyr::mutate(id = 1:n(),
        partisanship = as.factor(partisanship),
        skilled_labor = as.factor(skilled_labor),
        salary = as.factor(salary),
        benefits_bi = as.factor(benefits_bi),
        city = as.factor(city))

## Clean responses
df_cjoint_response_ven <- df_cjoint_ven %>%
 dplyr::select(id, matches("con[1-5]")) %>%
 gather(key = round, value = profile_choice, -id) %>%
 mutate(round = parse_number(round))

## Clean attributes
df_cjoint_covs_ven <- df_cjoint_ven %>%
 dplyr::select(id, starts_with("rd_")) %>%
 gather(key = feature_round_profile, value = attribute, -id) %>%
 mutate(round = parse_number(feature_round_profile),
     profile = case_when(grepl("_a_", feature_round_profile)~"a",
               grepl("_b_", feature_round_profile)~"b",
               TRUE~"ruh-roh"),
     feature = gsub("rd_[1-5]_[a|b]_", "", feature_round_profile)) %>%
 dplyr::select(-feature_round_profile) %>%
 spread(key = feature, value = attribute)

## Bind responses and attributes together
df_cjoint_clean_ven <- inner_join(df_cjoint_response_ven, df_cjoint_covs_ven,
               by = c("id", "round")) %>%
 left_join(df_cjoint_ven %>% dplyr::select(id,
                      partisanship,
                      skilled_labor,
                      salary,
                      benefits_bi,
                      city)) %>%
 mutate(chosen_profile = case_when(profile_choice == 1 & profile == "a"~1,
                  profile_choice == 2 & profile == "b"~1,
                  profile_choice == 1 & profile == "b"~0,
                  profile_choice == 2 & profile == "a"~0,
                  TRUE~NA_real_),
     PARTISANSHIP = as.factor(Partisanship), 
     `SKILL LEVEL` = as.factor(`Skill Level`), 
     `IDENTITY OF MIGRANT` = as.factor(`Identity of Migrant`),
     `PROBABILITY OF EMPLOYMENT` = as.factor(`Probability of Employment`), 
      RACE = as.factor(Race), 
     `REASON FOR LEAVING` = as.factor(`Reason for Leaving`), 
     GENDER = as.factor(Gender)
     ) %>%
 drop_na(-c(partisanship,
        skilled_labor,
        salary,
        benefits_bi,
        city)) %>%
 mutate(pair_id = paste0(id, round)) %>%
 group_by(pair_id) %>%
 filter(n() == 2) %>% ungroup()
 #drop_na(profile_choice)

## Run conjoint estimator
set_baseline <- list(PARTISANSHIP = "Left",
           `SKILL LEVEL` = "Low",
           `IDENTITY OF MIGRANT` = "Colombians displaced from around Cali",
           `PROBABILITY OF EMPLOYMENT` = "Cannot work",
            RACE = "Mestizo",
           `REASON FOR LEAVING` = "Fear of crime in their area")

df_cjoint_clean_ven$PARTISANSHIP <- fct_relevel(df_cjoint_clean_ven$PARTISANSHIP, "Left", "Center", "Right")
df_cjoint_clean_ven$`SKILL LEVEL` <- fct_relevel(df_cjoint_clean_ven$`SKILL LEVEL`, "Low", "Medium", "High")

amce_out.ven <- cjoint::amce(chosen_profile ~ PARTISANSHIP + `IDENTITY OF MIGRANT` + `SKILL LEVEL` +
    `PROBABILITY OF EMPLOYMENT` +`REASON FOR LEAVING` + RACE + GENDER,
   data = df_cjoint_clean_ven, 
   respondent.id = "id", 
   baselines = set_baseline)

amce_out.ven$estimates <- lapply(amce_out.ven$estimates, function(x){
 est <- x[1,]
 stderr <- x[2,]
 ci_low <- est - stderr * qnorm(.975)
 ci_hi <- est + stderr * qnorm(.975)
 xout <- rbind(est, stderr, ci_low, ci_hi)
 colnames(xout) <- colnames(x)
 return(xout)
})

## Plot results
cbPalette <- c("#99098d", "#000000", "#D55E00", "#009E73", "#E69F00", "#999999", "#0072B2")

amce_out.ven.plot <- cjoint_plot(amce_out.ven, 
   xlim=c(-.27,.18), breaks=c(-.2, -.1, 0, .1), labels=c("-.2", "-.1", "0",".1"), 
   xlab="Change in Pr(Preferred Migrant)",
   #main="Venezuelan respondents",
   lwd = 1.5,
   colors = cbPalette,
   group.order = c("PARTISANSHIP", 
                   "IDENTITY OF MIGRANT", 
                   "SKILL LEVEL",
                   "PROBABILITY OF EMPLOYMENT", 
                   "REASON FOR LEAVING", 
                   "RACE", 
                   "GENDER")) +
   theme_bw() + 
   theme(legend.position = "none") 

## plot
amce_out.col.plot

@

The conjoint experiment allows us to understand the role of political anxieties compared to other common concerns about migrant reception. Figure \ref{fig:main_conj} shows that migrants' political views are a major determinant in their reception.
Among Colombians, respondents preferred migrants who sympathized with the political center or the right: the AMCE for supporting the center was 
\Sexpr{formatC(amce_out.col$estimates$PARTISANSHIP[[1,1]], digits = 1)} 
(95\% CI =[\Sexpr{formatC(amce_out.col$estimates$PARTISANSHIP[[3,1]], digits = 1)},
\Sexpr{formatC(amce_out.col$estimates$PARTISANSHIP[[4,1]], digits = 1)}]) and for the right was 
\Sexpr{formatC(amce_out.col$estimates$PARTISANSHIP[[1,2]], digits = 1)} 
(95\% CI =[\Sexpr{formatC(amce_out.col$estimates$PARTISANSHIP[[3,2]], digits = 1)},
\Sexpr{formatC(amce_out.col$estimates$PARTISANSHIP[[4,2]], digits = 1)}]). 
The main penalty comes from holding leftist views; center and right-wing migrants gain similar levels of support. 
Answering this same conjoint exercise, shown in SI Figure \ref{fig:main_conj_ven}, Venezuelan respondents were also more supportive of migrants who sympathize with the center or right. This preference further reinforces how inaccurate Colombians' stereotypes of Venezuelans are in reality: most Venezuelan migrants don't want those who sympathize with the political Left either.

The experiment provides little support for alternative theories focused on migrants' labor market impacts or fiscal burden. Unlike similar conjoint experiments run in the US \citep{hainmueller2015hidden} or Europe \citep{bansak2016economic}, skill has a very small or null effect. Colombians actually slightly prefer migrants who are unlikely to find work and left due to poverty or violence, suggesting humanitarian concerns (see SI Section \ref{SIsec:conj} for additional analyses). The only other attribute with a large effect is the respondent's origin. We see strong national and regional preferences: respondents favor their co-nationals and co-regionals---a ``sons of the soil'' effect \citep{bhavnani2018nativism}. 


\section*{The Origins of Misperceptions}

%% PLOT OF MISPERCEPTIONS BY SUBGROUP
<<misperceptions_plot_subgroup, eval = TRUE, echo = FALSE, tidy=TRUE, fig.width = 8, fig.height = 3.5, out.width= "1\\linewidth", fig.pos='t', fig.align='center', warning=FALSE, message=FALSE, strip.white=TRUE, fig.cap="Comparing proportions of misperceptions about Venezuelans by respondent subgroups: whether they have direct contact with Venezuelans (black) and whether they live in Cúcuta vs Cali (blue dash).">>=

# BY CITY
df_city <- df_col %>%
  dplyr::select(c(pol_left, pol_maduro, pol_vote1, pol_vote2, votebuy_ven, pol_guerilla, sec_crime, city)) %>%
  pivot_longer(-city, names_to = "variable", values_to = "value") %>%
  drop_na() %>%
  split(.$variable) %>%
  map(~estimatr::difference_in_means(value ~ city, data = .x)) %>%
  map_dfr(~tidy(.x), .id = "outcome") %>%
  mutate(outcome = case_when(
    outcome == "pol_guerilla"~"Support guerrilla groups",
    outcome == "pol_left"~"Support the left",
    outcome == "pol_maduro"~"Support Maduro",
    outcome == "pol_vote1"~"Can vote in national elections",
    outcome == "pol_vote2"~"Can vote in local elections",
    outcome == "sec_crime"~"Have criminal ties to gangs",
    outcome == "votebuy_ven"~"Received a vote-buying offer"
  ))

# BY CONTACT
df_contact <- df_col %>%
  dplyr::select(c(pol_left, pol_maduro, pol_vote1, pol_vote2, votebuy_ven, pol_guerilla, sec_crime, dir_contact_bi)) %>%
  pivot_longer(-dir_contact_bi, names_to = "variable", values_to = "value") %>%
  drop_na() %>%
  split(.$variable) %>%
  map(~estimatr::difference_in_means(value ~ dir_contact_bi, data = .x)) %>%
  map_dfr(~tidy(.x), .id = "outcome") %>%
  mutate(outcome = case_when(
    outcome == "pol_guerilla"~"Support guerrilla groups",
    outcome == "pol_left"~"Support the left",
    outcome == "pol_maduro"~"Support Maduro",
    outcome == "pol_vote1"~"Can vote in national elections",
    outcome == "pol_vote2"~"Can vote in local elections",
    outcome == "sec_crime"~"Have criminal ties to gangs",
    outcome == "votebuy_ven"~"Received a vote-buying offer"
  ))

misperceptions_bycontactandcity <- bind_rows(
  df_contact %>% 
    mutate(comparison = "Direct Contact vs No Direct Contact"),
  df_city %>%
    mutate(comparison = "Cúcuta vs Cali")
) %>%
  mutate(comparison = fct_relevel(comparison, "Direct Contact vs No Direct Contact", "Cúcuta vs Cali")) %>%
  ggplot(aes(outcome, estimate, group = comparison, linetype = comparison,
             colour = comparison, label = round(estimate, 2))) + 
  geom_hline(aes(yintercept = 0), lty = "dashed") + 
  geom_errorbar(aes(ymin = conf.low, ymax = conf.high), width = 0, 
                position = position_dodge(.9)) + 
  geom_label(size=3, position = position_dodge(.9)) +
  scale_colour_manual(values=c("#000000", "#0072B2")) +
  scale_x_discrete(labels = function(x) str_wrap(x, width = 10)) +
  scale_y_continuous(limits = c(-.2, .25)) + 
  labs(x = "", y = "Difference in Proportion") +
  annotate(geom="text", x=1.7, y=.24, label="Direct Contact vs No Direct Contact", color="#000000") +
  annotate(geom="text", x=1, y=.2, label="Cúcuta vs Cali", color="#0072B2") +
  ggtitle("Subgroup Comparison of Misperceptions") +
  theme_bw() + 
  theme(legend.position = "none")

# plot
misperceptions_bycontactandcity

@

It is difficult to isolate the origins of political misperceptions, but we consider three possible channels: limited contact, welfare dependence, and electoral concerns.  First, we find that direct contact is not associated with more accurate perceptions. Figure \ref{fig:misperceptions_plot_subgroup} shows no differences in the proportion of respondents who have direct contact with Venezuelans versus those who do not on holding various misperceptions. The lack of difference is surprising given the potential for reverse causation in social interactions (i.e., Colombians who hold fewer misperceptions should be more willing to have direct contact with Venezuelans). Similarly, respondents in C\'ucuta may be expected to have more accurate views of Venezuelans, given a larger and longer history of cross-border flows, compared to Cali. Yet Figure \ref{fig:misperceptions_plot_subgroup} shows, if anything, a greater proportion of respondents in C\'ucuta hold misperceptions. 

Second, we consider if respondents cast Venezuelans as leftist because they see migrants as economically precarious. Such welfare reasoning suggests a positive correlation between political misperceptions (i.e. that the majority of Venezuelans are leftists) and beliefs that Venezuelan migrants make it more difficult to access social services and drive up Colombians' taxes. Yet SI Section \ref{SIsec:miswel} shows that there is no relationship between these questions. Furthermore, if Colombians are worried about Venezuelans' economic vulnerability, they should be less willing to accept unskilled or unemployed migrants.  Colombians instead somewhat prefer to admit more vulnerable migrants.  An even more precise prediction is that political ideology may matter more for economically vulnerable migrants.  But there is no interaction effect between the political views and the skill or employability of migrants (SI Section \ref{SIsec:acie}). 

Third, we consider evidence that individuals' misperceptions reflect electoral concerns.  The key empirical prediction is that individuals prefer migrants who share their ideological views.  The left panel of Figure \ref{fig:cond_conj} shows the conditional AMCE based on a respondent's own political ideology. As expected under an electoral theory, preferences for right-wing migrants are strongest for those who identify with the political Right.  Those on the Left also seem to prefer left-wing migrants, but the effects are imprecisely estimated due to the small number of left-leaning respondents.\footnote{Results are substantively the same with marginal means (SI Section \ref{SIsec:mm}).}

If electoral concerns motivate migrant receptions, then citizens might care most about migrants' politics when they misperceive their voting rights or live in an area where the electoral effects will be larger.  The center panel of Figure \ref{fig:cond_conj} conditions on whether the respondent misperceives that migrants can vote in national elections or not. The effects are correctly signed, but not statistically significant.  It is possible that Colombians worry about the future electoral impacts of migrants, so current misperceptions matter little.  We also examine whether citizens care more about ideological leanings where migrants have greater ``electoral leverage'' due to their group size and spatial concentration \citep{dancygier2017dilemmas}. However, political fears matter equally in Cali and C\'ucuta, as shown on the right panel, despite much larger Venezuelan concentrations in C\'ucuta.  A possible confounder is that respondents struggle to perceive electoral demographics: whereas Colombians in Cali report having fewer Venezuelan friends (a measure of direct, interpersonal contact), they report seeing Venezuelans in stores, public offices, and begging (indirect contact) as much as those in C\'ucuta do. Cali residents thus perceive that Venezuelans are equally present as in C\'ucuta, which may explain a shared sense of electoral threat.

<<cond_conj, eval = TRUE, echo = FALSE, tidy=TRUE, fig.width = 9.5, fig.height = 5, out.width= "1\\linewidth", fig.align='center', warning=FALSE, message=FALSE, strip.white=TRUE, fig.cap="AMCE estimates with 95\\% CIs of migrant profile political polarization, conditional on respondents' own political ideology (left), whether they have the misperception that Venezuelans can vote in the national elections (center), and respondents' city (right).">>=

#### Respondent political polarization

#### Colombians
amce_out.col <- cjoint::amce(chosen_profile ~ PARTISANSHIP:partisanship + `IDENTITY OF MIGRANT` + `SKILL LEVEL` +
       `PROBABILITY OF EMPLOYMENT` +`REASON FOR LEAVING` + RACE + GENDER,
     data = df_cjoint_clean_col, 
     respondent.id = "id", respondent.varying = "partisanship",
     baselines = set_baseline)

## Plot results
amce_out.col <- cjoint_plot(amce_out.col) 
amce_out.col <- amce_out.col$data %>% 
  filter(grepl("PARTISANSHIP", var)) %>%
  filter(printvar %in% c("   Center", "   Right")) %>%
  mutate(facet = as.character(facet),
         facet2 = case_when(facet == "Unconditional"~"Unconditional",
                           facet == "Conditional on\npartisanship = 1"~"Conditional on Left",
                           facet == "Conditional on\npartisanship = 2"~"Conditional on Center",
                           facet == "Conditional on\npartisanship = 3"~"Conditional on Right",
                           TRUE~"ruh-roh"),
         facet3 = fct_relevel(facet2, "Unconditional", 
                             "Conditional on Left", 
                             "Conditional on Center",
                             "Conditional on Right"),
         printvar = fct_relevel(printvar, 
                                "   Right",
                                "   Center"))

amce_out_pol.col.plot <- ggplot(amce_out.col, aes(printvar, pe)) + 
  geom_hline(aes(yintercept = 0), lty = "dashed") + 
  geom_point(size = 3, colour = "#0072B2") + 
  geom_errorbar(aes(ymin = lower, ymax = upper), 
                width = 0, size = .5, colour = "#0072B2") + 
  geom_label(aes(label = gsub("0\\.", "\\.", round(pe, 2))), size = 3, colour = "#0072B2") + 
  ylim(-.15, .2) +
  facet_wrap(~facet3, ncol = 1) + 
  #ggtitle("Colombian respondents") +
  ggtitle("By Respondent Ideology") +
  ylab("Change in Pr(Preferred Migrant)") +
  theme_bw() + 
  theme(legend.position = "none",
        axis.title.y = element_blank()) +
  coord_flip()

#### Venezuelans
amce_out.ven <- cjoint::amce(chosen_profile ~PARTISANSHIP:partisanship + `IDENTITY OF MIGRANT` + `SKILL LEVEL` +
       `PROBABILITY OF EMPLOYMENT` +`REASON FOR LEAVING` + RACE + GENDER,
     data = df_cjoint_clean_ven[!is.na(df_cjoint_clean_ven$partisanship),], 
     respondent.id = "id", respondent.varying = "partisanship",
     baselines = set_baseline)

## Plot results
amce_out.ven <- cjoint_plot(amce_out.ven) 
amce_out.ven <- amce_out.ven$data %>% 
  filter(grepl("PARTISANSHIP", var)) %>%
  filter(printvar %in% c("   Center", "   Right")) %>%
  mutate(facet = as.character(facet),
         facet2 = case_when(facet == "Unconditional"~"Unconditional",
                           facet == "Conditional on\npartisanship = 1"~"Conditional on Left",
                           facet == "Conditional on\npartisanship = 2"~"Conditional on Center",
                           facet == "Conditional on\npartisanship = 3"~"Conditional on Right",
                           TRUE~"ruh-roh"),
         facet3 = fct_relevel(facet2, "Unconditional", 
                             "Conditional on Left", 
                             "Conditional on Center",
                             "Conditional on Right"),
         printvar = fct_relevel(printvar, 
                                "   Right",
                                "   Center"))

amce_out_pol.ven.plot <- ggplot(amce_out.ven, aes(printvar, pe)) + 
  geom_hline(aes(yintercept = 0), lty = "dashed") + 
  geom_point(size = 3, colour = "#0072B2") + 
  geom_errorbar(aes(ymin = lower, ymax = upper), 
                width = 0, size = .5, colour = "#0072B2") + 
  geom_label(aes(label = gsub("0\\.", "\\.", round(pe, 2))), size = 3, colour = "#0072B2") + 
  ylim(-.15, .2) +
  facet_wrap(~facet3, ncol = 1) + 
  ggtitle("Venezuelan respondents") +
  ylab("Change in Pr(Preferred Migrant)") +
  theme_bw() + 
  theme(legend.position = "none",
        axis.title.y = element_blank(),
        axis.text.y = element_blank()
        ) +
  coord_flip()

#### Respondent misperception on national voting (COLs ONLY)

#### Colombians
amce_out.col <- cjoint::amce(chosen_profile ~ PARTISANSHIP:pol_vote1 + `IDENTITY OF MIGRANT` + `SKILL LEVEL` +
       `PROBABILITY OF EMPLOYMENT` +`REASON FOR LEAVING` + RACE + GENDER,
     data = df_cjoint_clean_col, 
     respondent.id = "id", respondent.varying = "pol_vote1",
     baselines = set_baseline)

## Plot results
amce_out.col <- cjoint_plot(amce_out.col) 
amce_out.col <- amce_out.col$data %>% 
  filter(grepl("PARTISANSHIP", var)) %>%
  filter(printvar %in% c("   Center", "   Right")) %>%
  mutate(facet = as.character(facet),
         facet2 = case_when(facet == "Unconditional"~"Unconditional",
                           facet == "Conditional on\npol_vote1 = 0"~"Conditional on No Voting Misperception",
                           facet == "Conditional on\npol_vote1 = 1"~"Conditional on Voting Misperception",
                           TRUE~"ruh-roh"),
         facet3 = fct_relevel(facet2, "Unconditional", 
                             "Conditional on No Voting Misperception", 
                             "Conditional on Voting Misperception"),
         printvar = fct_relevel(printvar, 
                                "   Right",
                                "   Center"))

amce_out_vote1.col.plot <- ggplot(amce_out.col, aes(printvar, pe)) + 
  geom_hline(aes(yintercept = 0), lty = "dashed") + 
  geom_point(size = 3, colour = "#0072B2") + 
  geom_errorbar(aes(ymin = lower, ymax = upper), 
                width = 0, size = .5, colour = "#0072B2") + 
  geom_label(aes(label = gsub("0\\.", "\\.", round(pe, 2))), size = 3, colour = "#0072B2") + 
  ylim(-.15, .2) +
  facet_wrap(~facet3, ncol = 1) + 
  #ggtitle("Colombian respondents") +
  ggtitle("By Respondent Misperception\non Venezuelan Access to Vote") +
  ylab("Change in Pr(Preferred Migrant)") +
  theme_bw() + 
  theme(legend.position = "none",
        axis.title.y = element_blank()) +
  coord_flip()

#### Respondent city

#### Colombians
amce_out.col <- cjoint::amce(chosen_profile ~ PARTISANSHIP:city + `IDENTITY OF MIGRANT` + `SKILL LEVEL` +
       `PROBABILITY OF EMPLOYMENT` +`REASON FOR LEAVING` + RACE + GENDER,
     data = df_cjoint_clean_col, 
     respondent.id = "id", respondent.varying = "city",
     baselines = set_baseline)

## Plot results
amce_out.col <- cjoint_plot(amce_out.col) 
amce_out.col <- amce_out.col$data %>% 
  filter(grepl("PARTISANSHIP", var)) %>%
  filter(printvar %in% c("   Center", "   Right")) %>%
  mutate(facet = as.character(facet),
         facet2 = case_when(facet == "Unconditional"~"Unconditional",
                           facet == "Conditional on\ncity = Cali"~"Conditional on Cali",
                           facet == "Conditional on\ncity = Cúcuta"~"Conditional on Cúcuta",
                           TRUE~"ruh-roh"),
         facet3 = fct_relevel(facet2, "Unconditional", 
                             "Conditional on Cali", 
                             "Conditional on Cúcuta"),
         printvar = fct_relevel(printvar, 
                                "   Right",
                                "   Center"))

amce_out_city.col.plot <- ggplot(amce_out.col, aes(printvar, pe)) + 
  geom_hline(aes(yintercept = 0), lty = "dashed") + 
  geom_point(size = 3, colour = "#0072B2") + 
  geom_errorbar(aes(ymin = lower, ymax = upper), 
                width = 0, size = .5, colour = "#0072B2") + 
  geom_label(aes(label = gsub("0\\.", "\\.", round(pe, 2))), size = 3, colour = "#0072B2") + 
  ylim(-.15, .2) +
  facet_wrap(~facet3, ncol = 1) + 
  ggtitle("By Respondent City") +
  #ggtitle("Colombian respondents") +
  ylab("Change in Pr(Preferred Migrant)") +
  theme_bw() + 
  theme(legend.position = "none",
        axis.title.y = element_blank()) +
  coord_flip()

#### Venezuelans
amce_out.ven <- cjoint::amce(chosen_profile ~ PARTISANSHIP:city + `IDENTITY OF MIGRANT` + `SKILL LEVEL` +
       `PROBABILITY OF EMPLOYMENT` +`REASON FOR LEAVING` + RACE + GENDER,
     data = df_cjoint_clean_ven, 
     respondent.id = "id", respondent.varying = "city",
     baselines = set_baseline)

## Plot results
amce_out.ven <- cjoint_plot(amce_out.ven) 
amce_out.ven <- amce_out.ven$data %>% 
  filter(grepl("PARTISANSHIP", var)) %>%
  filter(printvar %in% c("   Center", "   Right")) %>%
  mutate(facet = as.character(facet),
         facet2 = case_when(facet == "Unconditional"~"Unconditional",
                           facet == "Conditional on\ncity = Cali"~"Conditional on Cali",
                           facet == "Conditional on\ncity = Cúcuta"~"Conditional on Cúcuta",
                           TRUE~"ruh-roh"),
         facet3 = fct_relevel(facet2, "Unconditional", 
                             "Conditional on Cali", 
                             "Conditional on Cúcuta"),
         printvar = fct_relevel(printvar, 
                                "   Right",
                                "   Center"))

amce_out_city.ven.plot <- ggplot(amce_out.ven, aes(printvar, pe)) + 
  geom_hline(aes(yintercept = 0), lty = "dashed") + 
  geom_point(size = 3, colour = "#0072B2") + 
  geom_errorbar(aes(ymin = lower, ymax = upper), 
                width = 0, size = .5, colour = "#0072B2") + 
  geom_label(aes(label = gsub("0\\.", "\\.", round(pe, 2))), size = 3, colour = "#0072B2") + 
  ylim(-.15, .2) +
  facet_wrap(~facet3, ncol = 1) + 
  ggtitle("Venezuelan respondents") +
  ylab("Change in Pr(Preferred Migrant)") +
  theme_bw() + 
  theme(legend.position = "none",
        axis.title.y = element_blank(),
        axis.text.y = element_blank()
        ) +
  coord_flip()


## plot
amce_out_pol.col.plot + amce_out_vote1.col.plot + amce_out_city.col.plot + plot_layout(ncol=3, widths=c(1,1,1))

@

Finally, we consider whether political elites foster misperceptions.  We look for evidence that national politicians spread false information about Venezuelan migrants' political incorporation and that these rumors diffused across the country.\footnote{We cannot directly test the effect of these rumors as they were ubiquitous across traditional media and social media like WhatsApp.} Rumors spread on social and news media that Venezuelans living in Colombia would swing the election. One widely shared post claimed that the Colombian government expanded Venezuelans' residency permits so they could vote in the presidential elections.\footnote{``Es falso que est\'an ofreciendo nacionalidad colombiana a los venezolanos para que voten en las elecciones,'' Colombia Check, July 18, 2019.} Another viral audio clip featured a supposed Venezuelan leader assuring a Colombian woman, ``[Venezuelans] have only come to register as voters, and they are going to support Colombia voting for Petro, for all that is communism.''\footnote{``Campa\~{n}a sucia? Cadena de WhatsApp advierte de venezolanos registrados para votar por Presidente,'' La FM Radio, February 28, 2018, \url{www.lafm.com.co/politica/campana-sucia-cadena-de-whatsapp-advierte-de-venezolanos-registrados-para-votar-por-presidente}} Stories in major newspapers and informal WhatsApp message groups that often got picked up local radio stations emphasized that Venezuelans sympathize with left-wing leaders like Hugo Ch\'avez, as well as Fidel Castro, and would ``infect'' Colombian society \citep{Ordonez2019}. Beyond these rumors, former President Uribe tweeted constantly to his five million Twitter followers, who may have read and spread his messages warning against a ``second Venezuela'' and the specter of ``castrochavismo'' (see SI Section \ref{SIsec:tweets} for examples). 

These messages were repeated by politicians prior to regional and local elections in 2019. Campaign slogans for one of the main right-wing political parties, Centro Democr\'atico, in Bogot\'a, Bucaramanga, and Cali included, ``I don't want to live like a Venezuelan,'' and ``So Colombia won't be another Venezuela'' \citep{Ordonez2019}. Rumors also circulated that politicians granted Venezuelans identification documents in exchange for campaigning and voting for them. In Cali, the mayor, Maurice Armitage, allegedly distributed ID cards to 11,000 Venezuelans for participating in the campaign of his ally, Alejandro \'Eder.\footnote{``No, en Cali no est\'an dando c\'edulas de extranjer\'ia a venezolanos a cambio de votos,'' Colombia Check, December 3, 2018.}  In short, the findings are most consistent with a theory in which citizens care about the electoral incorporation of migrants with opposing political views, and political elites mischaracterize migrants' views to advance their support.


\section*{Conclusion}

In much of the Global North, differences in ethnic, racial, or religious backgrounds overlap with perceived differences in political views. In this article, we studied a context in which migrants are similar in demographics but come from a country that followed a divergent political trajectory. We showed how Colombian politicians, media outlets, and citizens constructed an out-group based on Venezuelans' political leanings. Right-wing politicians, in particular, have underscored a socialist threat from Venezuelan migrants. This article provides evidence that these elite messages resonated among local communities.  Host communities worry about the electoral impacts of migration, particularly when the perceived political leanings of migrants cut against their own preferences.

We used original survey data to show a substantial gap between Colombians' perceptions and Venezuelan migrants' reported political views. While Venezuelan respondents were more right-wing than Colombians, many Colombians perceived them as left-wing. In a conjoint experiment, respondents showed a strong bias against left-leaning migrants, especially when they themselves held right-wing political views. Surprisingly, political views played a more important role than race, skill, or poverty in shaping migrant reception. 

While we find robust support for a strategic electoral theory, reliance on a conjoint experiment in a single country raise concerns about generalizability. First, the survey took place around local elections where candidates have greater incentives to highlight the political impacts of migration and voters are focused on politics. A related issue is that a forced choice conjoint requires respondents to pick a migrant, whereas ``true'' preferences may involve accepting no migrants.  Future research should look at the role of political fears in between electoral cycles and the rejection of all migration.  But even if political fears fade and lead to a dislike of all migrants, they can be an important determinant of public attitudes towards migrants as elections approach.  

Second, Colombia has received more and poorer migrants than the rest of Latin America because migrants can reach the border by foot.  Political fears may operate differently in countries that receive somewhat wealthier migrants. Yet, if anything, wealthier Venezuelans--who have the capital to reach more distant destinations--are more likely to be right-wing regime opponents.  While no surveys exist to judge political misperceptions across Latin America, politicians have tried to associate Venezuelan migrants with left-wing threats in elections in Brazil, Peru, and even the United States.  Our findings likely apply best to other countries still shaped by post-Cold War cleavages where voters (even those with little contact with migrants) are wary of the extreme political ideologies that migrants tend to flee.  

An important implication of these findings is that the political consequences of migration are likely to be the opposite of those feared by host countries. As Venezuelans gain the right to vote, they are likely to push the electorate further right. Indeed, consistent with our findings, one study of municipal election results shows an increase in support for the political right where Venezuelan migration has increased the most \citep{rozo2021brothers}. Yet the channel of electoral impact differs from standard assumptions that the Right gains due to growing xenophobia.  We suggest that voters turn to the political right out of fear of following a far leftist path like Venezuela. This trend may only accelerate when migrants, who overwhelmingly sympathize with the Right, soon gain the right to vote in local elections. Our results also reinforce emigration's potential impact on politics in Venezuela: even those fleeing for economic reasons are deeply dissatisfied with Maduro. Migrants may have strengthened the political opposition to the regime had they stayed.

Given the gap between perceived and actual political views, our results suggest a potential role for the media to strengthen support for migrants and for future research on how political parties design their migration platforms. Rumors circulating through social media about Venezuelans can create antipathy among Colombians, particularly right-leaning Colombians. Campaigns providing accurate information about whether migrants can participate in elections and Venezuelans' critiques of their government's policies could potentially help build support for their resettlement. Research remains to be done on how host communities learn about migrants' political views and whether media and political campaigns can correct misperceptions.  

One particularly interesting extension would be to study the formation of political misperceptions in a context where migrants flee a right-wing government. We suspect that receiving communities would falsely associate migrants with the government that they flee, in that case, the political Right. In contrast, theories focused on economic vulnerability still would predict that migrants are stereotyped as leftists. Extending our original methods of comparing the perceptions of host communities and migrants to additional contexts with different ideological cleavages would build an even more robust theory of political misperceptions and their consequences.    

Finally, this article suggests a tension for leaders of conservative political parties. While the political Right often is associated with anti-immigrant messages, immigrants may become an important bastion of political support in contexts where they flee left-wing regimes. How right-wing parties campaign to attract anti-immigrant voters while courting migrant votes is an interesting dilemma. Whether migrants are portrayed as ``unskilled,'' ``culturally backward,'' ``leftist,'' or ``terrorists'' depends on how parties create cleavages to boost their electoral prospects. 


%%%%%% ACKNOWLEDGEMENTS %%%%%%%
\newpage
\noindent\textbf{Acknowledgments}: We are grateful to colleagues at the Centro Nacional de Consultoria (CNC) for survey administration, and to Olivia Woldemikael and Daniel Rojas Lozano for excellent research assistance.  We thank Fiona Adamson, Hannah Alarian, Edward Gonzalez, Guy Grossman, Nicol\'as Idrobo, Patrick James, Efr\'en P\'erez, and participants from the Global Research in International Political Economy (GRIPE) seminar, the UCLA CSIM Seminar, APSA 2020, USC CIS International Relations Workshop, and the University of Pennsylvania's PDRI Barriers and Bridges to Immigrants' Integration Conference for helpful comments.


%%%%%% REFERENCES %%%%%%%
\newpage
\bibliography{colombia}


%%%%%% BIOGRAPHICAL STATEMENTS%%%%%%%
\newpage
\noindent\textbf{Biographical Statements}:\\
Alisha Holland is an Associate Professor of Government at Harvard University, Cambridge, MA 02138.\\
Margaret E. Peters is a Professor of Political Science Chair of the Global Studies Interdisciplinary Program at UCLA, Los Angeles, CA 90095.\\
Yang-Yang Zhou is an Assistant Professor of Government at Dartmouth College, Hanover, NH 03755.


\end{document}

